gwyn@Brl@sri-unix (09/02/82)
From: Doug Gwyn <gwyn@Brl> Date: 29 Aug 82 2:47:02-EDT (Sun) In the June 1982 DECUS SS&OS SIG newsletter, "The Toolkit", there is an article by Norman Wilson of Caltech describing their paging VAX UNIX/TS. Part of this article is a criticism (valid, I think) of Berkeley's way of hacking UNIX.
gwyn@Brl@sri-unix (09/05/82)
From: Doug Gwyn <gwyn@Brl> Date: 2 Sep 82 8:09:29-EDT (Thu) I appreciate your taking the time to discuss the evolutionary aspects of UNIX. Unfortunately we are faced with two UNIX families now, although UCB says that 4.2bsd will have System III compatibility (hope it does). My personal complaint with 4.1bsd has been the undue emphasis on efficiency issues to the detriment of compatibility with the Bell System UNIXes (which we will continue to be stuck with, like it or not). This wouldn't perhaps have been so bad if it weren't for these publicly-funded gurus stating repeatedly that one should toss his PDP-11 and get a VAX (words to that effect). The small, elegant approach to system design that attracted many of us to UNIX in the first place was lost in much of 4.1bsd, in my opinion. I believe the famous VMS-vs-UNIX debate may have resulted in UCB's trying to prove that UNIX can be as "efficient" as VMS, but at what cost? By the way, I have no major complaints about UNIX 3.0; it feels solid to use and really hasn't been badly damaged inside. The FIFOs integrated naturally into the existing pipe mechanism, so I can't fault USG for providing them while we await development of the perfect IPC mechanism.
obrien (09/10/82)
It would be well to remember that Berkeley is being funded by ARPA to provide a version of UNIX that meets the needs of a particular community: Image Processing and VLSI Design. These people need support for VERY large addressing spaces and GREAT efficiency. All non-portable work has been in support of these goals. (I am not an official Berkeley spokesman here, but this is what I understand.) They would be altogether happy if some company would pick up the ball and support 4.2 for everyone else, and fold in features, enhancements, portability, etc. that other folks want. They have just enough manpower to do what they're paid to, however. To the extent that they can put their system together in such a way as to benefit others, well and good, and they've done a tremendous job here. However, the enhancements which seem of dubious value to many have in fact been dictated by the environment for which the system is destined.